Abstract:
Linux kernel maintainers recently reached a consensus and decided to cancel the requirement that developers disclose the use of artificial intelligence-assisted programming tools. With the "Assisted-by" tag officially abolished, developers who submit code to the Linux kernel in the future do not need to specifically state it in their submission records even if they use large language models or other AI tools.
This decision originated from the Linux Maintainers Summit held in Prague, Czech Republic on October 8, 2026. The summit ultimately agreed to scrap the “Assisted-by” label and remove the mandatory requirement from related documents. Previously, developers who used AI to generate or assist in generating code were required to indicate the AI tool used in their submission information to explain that the relevant code was not entirely written by humans.
After the rule adjustment, developers submitting AI-assisted generated code will no longer need to make additional declarations, which means that the Linux kernel project has officially abandoned the mandatory disclosure mechanism for the use of AI tools.
Revolving around this change, Linux founder Linus Torvalds’ statement on the mailing list attracted widespread attention. When some developers worried that removing the disclosure requirement would weaken the project's oversight of AI code, Torvalds made it clear that arguments against AI tools were "extremely stupid." He believes that the key to the problem is not what tools developers use, but whether they truly understand the code they submit.
To illustrate his point, Torvalds gave an example. If developers are required not to use AI tools, then according to the same logic, developers can also be required to abandon compilers and write code directly in assembly language to prove that they truly understand the operating mechanism of the computer. He further quipped that even using a text editor to write commit instructions could be banned because the backspace key would mask typing errors. Torvalds believes that such rules are inherently meaningless.
Torvalds emphasized that what he really cares about is that contributors understand what the code is actually doing, not what tool the code was generated by. In his view, the tool itself does not determine the quality of the code, but the developer's understanding and sense of responsibility are the core criteria.
In fact, Torvalds has publicly expressed a positive attitude towards AI-assisted programming many times in recent months. He believes that AI programming tools will not only exist for a long time, but also improve development efficiency and even make the software development process more interesting. He himself has previously admitted that he has used AI to help understand complex patches, and in a recent kernel optimization work, he used Anthropic's Claude Code to generate implementation code, test content and initial documents, while the final review and performance verification work was completed by himself.
Those who support the cancellation of disclosure requirements believe that the enforcement effect of existing rules is limited, and some developers will not proactively mark the use of AI, so the relevant requirements lack enforceability. In addition, some maintainers pointed out that this label is equivalent to providing free publicity for commercial AI service providers, and the project itself has no clear reason to collect this information.
However, some developers have expressed concerns about the new policy. They believe that even if the rules cannot be enforced 100%, disclosure requirements can still convey the project's attitude and behavioral expectations to contributors and provide additional reference information for code reviewers. After removing the label, it may be harder to know whether certain codes were generated with the help of AI in the future.
As the "Assisted-by" label gradually withdraws from the historical stage, the Linux kernel project's position on the issue of AI programming tools has become more clear: the project will not focus on whether the code is written by AI, but will focus on code quality, functional correctness, and developers' understanding of the submitted content. This also makes Linux one of the most open representatives for AI-assisted development among current large-scale open source projects.
Related articles:
Linus Torvalds uses Claude Code to optimize the Linux kernel. Multi-threaded file reading performance is expected to be greatly improved
Linus Torvalds uses AI to reconstruct the readability of Linux kernel code and accidentally discovers three vulnerabilities that have existed for more than ten years
Comments